Thursday, September 26, 2013
The process of multiplexing is carried out at the transport layer. Several conversations are multiplexed in to one connection or physical links or virtual circuit. For example, suppose the host has only one network address available for use. Then it has to be used by all the transport connections originating at that host. For multiplexing the following two main strategies are followed:
Ø Upward multiplexing and
Ø Downward multiplexing
- In upward multiplexing, the different transport connections are multiplexed in to one network connection.
- These transport connections are grouped by the transport layer as per their destinations.
- It then maps the groups with the minimum number of network connections possible.
- The upward multiplexing is quite useful where the network connections come very expensive.
- It is only used when the connections with high bandwidth are required.
- In case of the downward multiplexing, the multiple network connections are opened by the transport layer and the traffic is distributed among them.
- But for using downward multiplexing, it is necessary that this capacity must be handled well by the subnet’s data links.
- In either of the cases it is not guaranteed that the segments will be delivered in order.
- Therefore, another technique is adopted.
- The segments are numbered sequentially.
- Each octet is numbered by the TCP sequentially.
- Segments are then numbered based up on the number of the first octet present in that segment.
- The segments might get damaged in the transition or some may even fail to arrive at the destination.
- This failure is not acknowledged by the transmitter.
- However, the successful receipt of the segment is does acknowledged by the receiver.
- Sometimes, the cumulative acknowledgements might be used.
- If the ACK triggers a time out interrupt, the re-transmission of the segment is done.
- Also the re-transmission is done when an ACK is lost.
- The receiver must have the ability to recognize the duplicate ACKs.
- If such thing occurs, the receiver assumes by itself the ACK might have been lost.
- This happens when the ACK duplicate is received before the connection is closed.
- If the duplicate is received after the closure of the connection, the situation is dealt differently.
- In this case, the sender and receiver are allowed to know about each other’s existence.
- They negotiate about the parameters and the transport entity resources are allocated based up on some mutual agreement.
The connection release is of two types:
Ø Asymmetric release:
This is the one used in the telephone systems. However it does not works well for the network that use packet switching.
Ø Symmetric release:
- This is certainly better than the previous one.
- Here, all the directions are released independently with respect to each other.
- The host continues receiving data after the disconnection TPDU has been sent.
- But the symmetric release has another problem which is related with indirection levels and fake messages.
- There are no proper solutions for this problem in case of the unreliable communication media.
- Note that this has nothing to do with the protocol.
- Putting a reliable protocol over an unreliable medium can actually guarantee the delivery of the message.
- Another thing to be noted is that it the time limit within which the message will be delivered cannot be guaranteed by any protocol.
- Error conditions might prolong the delivery period.
- Restarting the connections can lead to the loss of all the state info and the connection might remain as half-open.
- Since no protocol has been designed to deal with this problem therefore one has to go forward with the risks associated with releasing the connections.